Home / MCP / Vibe-Coder MCP Server

Vibe-Coder MCP Server

Provides a structured feature workflow with PRD/implementation planning, phased development, and document storage for LLM-driven coding features.

typescript
Installation
Add the following to your MCP client configuration file.

Configuration

View docs
{
    "mcpServers": {
        "vibe_coder_mcp": {
            "command": "/path/to/vibe-coder-mcp/build/mcp-server.js",
            "args": []
        }
    }
}

Vibe-Coder is a Model Context Protocol MCP server that guides LLM-based coding work through a structured feature clarification flow, PRD and implementation plan generation, phased development with task tracking, and built‑in document storage. It helps you organize feature work from concept to completion with clear progress reporting and accessible documents.

How to use

You use Vibe-Coder with an MCP client to plan, clarify, and track feature development. Start a feature clarification to gather requirements, then generate a PRD and implementation plan. Break the work into development phases and tasks, monitor progress, and keep feature documents stored and retrievable. Use the provided prompts and tools to guide questions, capture decisions, and store generated documents for later reference.

Typical usage flow: initiate a featureClarification to define intent and constraints, answer guided questions, request a PRD and implementation plan, create development phases, add tasks to each phase, and update statuses as work progresses. You can retrieve the location of generated documents and save copies to custom paths when needed.

How to install

Prerequisites: you need Node.js and npm installed on your machine. You also need a working MCP client that can connect to an MCP server via stdio.

Install dependencies for the MCP server project:

npm install

Build the server so it is ready to run in production or development environments:

npm run build

For development with automatic rebuilds, you can watch for changes and rebuild automatically:

npm run watch

Additional sections

Configuration and storage are designed to be practical and transparent. Generated documents (PRDs and implementation plans) are automatically saved to disk and kept in memory for quick access. By default, documents live under documents/{featureId}/ with filenames based on the document type.

Default storage locations and filenames:

  • documents/{featureId}/prd.md
  • documents/{featureId}/implementation-plan.md

You can save documents to custom locations using the saveDocument operation, and you can retrieve a document’s path using getDocumentPath to verify where a file lives and whether it has been saved.

{
  "featureId": "feature-123",
  "documentType": "prd",
  "filePath": "/custom/path/feature-123-prd.md"
}

Examples and troubleshooting

Debugging MCP servers can be challenging since they communicate over stdio. If you need debugging tools, use the MCP Inspector to access debugging utilities in your browser.

If you encounter issues during installation or runtime, verify you have the required Node.js version, check that dependencies install correctly, and ensure your development environment has access to the necessary file paths and permissions.

MCP connection configuration

The MCP server can be connected to via a local stdio transport. Use a configuration that points to the built server for local runs.

{
  "mcpServers": {
    "vibe_coder_mcp": {
      "type": "stdio",
      "command": "/path/to/vibe-coder-mcp/build/mcp-server.js",
      "args": []
    }
  }
}

Available tools

start_feature_clarification

Begin the feature clarification process to collect requirements, target users, success criteria, constraints, and success metrics.

provide_clarification

Answer clarification questions about a feature to refine its scope and requirements.

generate_prd

Generate a Product Requirements Document that captures the clarified feature details and acceptance criteria.

create_phase

Create a development phase for a feature to organize work into logical stages with clear goals.

add_task

Add specific tasks to a development phase to break work into actionable steps.

update_phase_status

Update the status of a development phase as work progresses.

update_task_status

Update the completion status of individual tasks.

get_next_phase_action

Get guidance on what to do next in the workflow based on current progress.

get_document_path

Get the path of a generated document and whether it has been saved to disk.

save_document

Save a generated document to a specified location.